CALIBRATION MANDATORY
Every prediction made anywhere in the Borg is logged via prediction-log. Brier scores are computed when outcomes resolve. Council reputation weights are recomputed when Brier shifts beyond threshold.
Overconfidence — the loudest councils become the most trusted regardless of accuracy
Confident predictions that turn out wrong are how all forecasting systems fail. Without mandatory calibration, the loudest councils become the most trusted regardless of accuracy. The Borg cannot afford this because the Borg's compounding advantage depends on getting better at prediction over time — which is only possible if predictions are scored honestly.
Every prediction routes through the prediction-log skill with three-lens architecture (Outside / Inside / Market View where applicable). Callback dates are stored. When the callback date arrives, the calibration-tracker skill prompts for outcome and computes Brier. Council reputation weights are recomputed if any lens Brier shifts >0.03 absolute.
Amendments to Law V would require demonstrating that calibration discipline is harmful to a specific council type. This is a high bar. The default position is that calibration applies universally.
Some councils don't naturally make predictions (Eulogists logs kills, Historians chronicles). Law V applies only to predictive claims — not all council outputs. The Bankers Council uses Brier scores to weight which councils get more Opus budget, which creates the right structural incentive.
Examples of what would violate Law V: a council making confident claims about future outcomes without logging them. Retroactive rationalisation of a missed prediction. Selective Brier reporting (only counting the wins). Any pattern where overconfidence is rewarded structurally.
Constitutional laws are not values — they are failure-mode insurance. Law V insures specifically against overconfidence. Without it, the Borg eventually fails through this specific mode regardless of how good the founder's intentions are.